1. Technical Field
The present invention relates generally to the field of voice signal compression in telecommunications networks to enhance transmission capacity and, more particularly, to the field of providing voice compression in existing telephone circuit switched networks that employ T1 and E1 frame formats while keeping the network infrastructure unchanged.
2. Description of the Relevant Art
Pulse code modulation for sampling voice signals and modulating a pulse coded data stream for transmission has been known since the 1960's. Two forms of pulse code modulation are the so-called .mu.-Law and the A-law modulation formats of T1 and E1 frames respectively. Both share the common principle that 8 bit pulse code words describe a speech signal or, alternatively, carry data or facsimile. In the T1 frame, 24 such eight bit words and a framing bit comprise a 193 bit frame. Each eight bit word describes a voice signal sample of a different speech communication. The eight bit words are formed into the 193 bit frame of FIG. 1(a) such that a framing bit 101 signals the beginning of the frame and/or is used for synchronization. The framing bit 101 is followed successively by the 24 8-bit .mu.-Law pulse coded words representing samples of 24 different voice communications or facsimile/data channels. The 24 words each represent a time slot or channel where timeslot or channel #1 is timeslot 102. Thus, for example, a maximum of twenty-four voice communications can be transmitted by one so-called DS1 channel bank. Timeslot or channel #2 is timeslot 103 and so on until the twenty-fourth time slot or channel #24 is represented as timeslot 104. Channels #3-23 are also timeslots and are indicated by the dotted box between timeslot 103 and timeslot 104.
The pulse code modulation process for encoding voice signals is well known. A speech wave is sampled at periodic discrete points in time to obtain pulses having different amplitudes. The speech signal amplitude is then quantized among, for example, 128 or 256 different levels and the least significant digit in each eight bit word in one frame out of six may be used for in band signaling. To quantize 256 levels requires 8 bits and, if the sampling rate is 8000 samples per second, the bit rate or information carrying capacity of each T1 carrier channel is 8 bits .times.8000 samples per second or 64 kbits/sec. In band signaling means carrying the signaling information for, for example, addressing or control information within the band of the T signal format. Out of band signaling recently has become preferred as an alternative or in addition to in band signaling where signaling information is transmitted via a separate transmission path, for example, via so-called SS-7 out-of-band signaling equipment. Referring briefly to FIG. 4, SS7 out-of-band signaling links are shown by dashed lines 480-487.
Referring to FIG. 1(b), there is shown a typical E1 frame data format wherein, instead of twenty-four channels or timeslots, thirty-two channels or timeslots are provided. The form of pulse code modulation is known as A-law pulse code modulation in the E-1 format. Of the thirty-two channels provided, thirty are utilized for carrying communications such as voice, fax and data communications. Timeslot or channel #1 is shown as timeslot 121; timeslot or channel #2 is shown as timeslot 122 and timeslot or channel #15 is shown as timeslot 123. Intermediate channels #3-14 which are also timeslots are indicated by the dotted omission. The sixteenth timeslot or channel, Timeslot #16, shown as timeslot 124, comprises 8 bits of in-band signaling data. Timeslot #17 or timeslot 125 is again a voice, data or fax channel. Timeslots #18-30 are not shown but as the dotted line omission, and Timeslot #31 or timeslot 126 is another voice, data or fax channel. Timeslot #32 comprises a predetermined eight bit framing signal 127.
Telecommunications traffic is carried on trunks between telephone switching offices. There are generally two types of telephone switching offices, a local switch and a toll or tandem switch. The local switch connects a telephone subscriber to the public switched telephone network. A tandem switch connects local switches or a local switch to a toll switch. A toll switch connects tandem switches to toll switches or connects toll switches. Trunks are sized traditionally into trunk groups based on the amount of traffic carried. A trunk that may have a capacity of 64 Kbits per second sits idle during non-peak periods and at busy periods wastes a portion of its 64 Kbits per second capacity carrying speech traffic.
Data and fax communications are presumed relatively data efficient in comparison with voice communications. Voice communications are frequented by periods of silence when no intelligible sounds, detectable as speech energy, are present. During a typical voice communication between parties talking together over a communications link, there are frequent periods of silence. Consequently, there is an opportunity in a voice communication to provide voice compression; that is, provide for utilization of periods of silence among other compression principles during the bandwidth of a voice communication by filling the silence with periods of voice from other communications. Both analog and digital forms of voice compression are known. Most, if not all, forms of voice compression utilize the dead or silence periods in speech to advantage. For example, a particular given period of time within a single voice communication channel may comprise a plurality of segments of speech from a related plurality of voice communications. In this manner, not just one voice communication is carried on the channel but a substantial increase in the number of concurrently handled calls on the same channel is obtained. The given period is broken into time slots and each time slot may comprise an active voice segment. Periods of silence are eliminated. A minor disadvantage is that the decompression and reassembly of the original voice communications carried over such a channel may take some time and so result in some delay, but the delay is not significant. Also, control information is required to describe the process of compression so that decompression can occur at a receiver. These are minor disadvantages in comparison to the enhancement in transmission capacity obtained. Moreover, practically none of the original speech content of the voice communication is lost. One known voice compression algorithm is that presently contemplated for application with video signal compression and is known as the M.P.E.G. II algorithm proposed standard.
The T carrier channel or timeslot is inherently inefficient, for example, timeslot 102, because the timeslot frequently carries periods of silence, silence that could be filled by voice segments of other voice communications. The E1 frame format is inefficient for the same reasons. Once a voice communications channel in either is dedicated to a particular voice communication in a call between two or more speaking parties, the channel remains so dedicated. There is no opportunity to share the voice communications channel. Of course, a fast talker makes more efficient use of the dedicated channel than a slow talker. Nevertheless, with either kind of caller, there is considerable inefficiency in communication.
To be competitive in today's telephony business, reducing the cost of handling telephone calls and increasing the existing network capacity have become crucial issues. Increasing the capacity of the network means the addition of more trunk facilities and network switches. However, this is a very expensive venture. Currently, a voice channel is transmitted at 64 Kbps in A-Law or .mu.-Law PCM format as described above with reference to FIG. 1. Dedication of a whole timeslot or 8 bit word as described above to voice is very expensive in terms of bandwidth utilization. Fax and data are transmitted in 64 Kbps bursts and so are more bandwidth efficient than voice. The existing T1 or E1 networks use T1 or E1 frames which contain twenty-four or thirty 64 Kbps voice channels, respectively. Each 64 Kbps voice channel or timeslot, contains one 8 bit word per T1 or E1 frame. The sampling rate is 8000 times per second. Since eight thousand frames are transmitted per second, the twenty-four channel bit rate is 24 channels .times.64 kbps per channel or 1.544 megabits per second including framing. The information transmission efficiency of this 1.544 megabit per second signal is much less. According to the well known digital multiplex hierarchy for digital data transmission, there is ample opportunity to improve the information carrying capacity at all levels from the so-called DS1 to DS4 levels and beyond. Consequently, it is an object of the present invention to improve the information carrying capacity of digital transmission facilities.
With the emergence of toll quality, low-bit rate speech coders and high-speed Digital Signal Processors (DSPs), an object of the present invention is to increase the network capacity by reducing the bandwidth of the voice channel and at the same time to maintain the voice signal at toll quality.